1 //===--- Format.cpp - Format C++ code -------------------------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 /// 10 /// \file 11 /// \brief This file implements functions declared in Format.h. This will be 12 /// split into separate files as we go. 13 /// 14 //===----------------------------------------------------------------------===// 15 16 #define DEBUG_TYPE "format-formatter" 17 18 #include "ContinuationIndenter.h" 19 #include "TokenAnnotator.h" 20 #include "UnwrappedLineParser.h" 21 #include "WhitespaceManager.h" 22 #include "clang/Basic/Diagnostic.h" 23 #include "clang/Basic/SourceManager.h" 24 #include "clang/Format/Format.h" 25 #include "clang/Lex/Lexer.h" 26 #include "llvm/ADT/STLExtras.h" 27 #include "llvm/Support/Allocator.h" 28 #include "llvm/Support/Debug.h" 29 #include "llvm/Support/YAMLTraits.h" 30 #include <queue> 31 #include <string> 32 33 namespace llvm { 34 namespace yaml { 35 template <> 36 struct ScalarEnumerationTraits<clang::format::FormatStyle::LanguageStandard> { 37 static void enumeration(IO &IO, 38 clang::format::FormatStyle::LanguageStandard &Value) { 39 IO.enumCase(Value, "Cpp03", clang::format::FormatStyle::LS_Cpp03); 40 IO.enumCase(Value, "C++03", clang::format::FormatStyle::LS_Cpp03); 41 IO.enumCase(Value, "Cpp11", clang::format::FormatStyle::LS_Cpp11); 42 IO.enumCase(Value, "C++11", clang::format::FormatStyle::LS_Cpp11); 43 IO.enumCase(Value, "Auto", clang::format::FormatStyle::LS_Auto); 44 } 45 }; 46 47 template <> 48 struct ScalarEnumerationTraits<clang::format::FormatStyle::BraceBreakingStyle> { 49 static void 50 enumeration(IO &IO, clang::format::FormatStyle::BraceBreakingStyle &Value) { 51 IO.enumCase(Value, "Attach", clang::format::FormatStyle::BS_Attach); 52 IO.enumCase(Value, "Linux", clang::format::FormatStyle::BS_Linux); 53 IO.enumCase(Value, "Stroustrup", clang::format::FormatStyle::BS_Stroustrup); 54 IO.enumCase(Value, "Allman", clang::format::FormatStyle::BS_Allman); 55 } 56 }; 57 58 template <> 59 struct ScalarEnumerationTraits< 60 clang::format::FormatStyle::NamespaceIndentationKind> { 61 static void 62 enumeration(IO &IO, 63 clang::format::FormatStyle::NamespaceIndentationKind &Value) { 64 IO.enumCase(Value, "None", clang::format::FormatStyle::NI_None); 65 IO.enumCase(Value, "Inner", clang::format::FormatStyle::NI_Inner); 66 IO.enumCase(Value, "All", clang::format::FormatStyle::NI_All); 67 } 68 }; 69 70 template <> struct MappingTraits<clang::format::FormatStyle> { 71 static void mapping(llvm::yaml::IO &IO, clang::format::FormatStyle &Style) { 72 if (IO.outputting()) { 73 StringRef StylesArray[] = { "LLVM", "Google", "Chromium", 74 "Mozilla", "WebKit" }; 75 ArrayRef<StringRef> Styles(StylesArray); 76 for (size_t i = 0, e = Styles.size(); i < e; ++i) { 77 StringRef StyleName(Styles[i]); 78 clang::format::FormatStyle PredefinedStyle; 79 if (clang::format::getPredefinedStyle(StyleName, &PredefinedStyle) && 80 Style == PredefinedStyle) { 81 IO.mapOptional("# BasedOnStyle", StyleName); 82 break; 83 } 84 } 85 } else { 86 StringRef BasedOnStyle; 87 IO.mapOptional("BasedOnStyle", BasedOnStyle); 88 if (!BasedOnStyle.empty()) 89 if (!clang::format::getPredefinedStyle(BasedOnStyle, &Style)) { 90 IO.setError(Twine("Unknown value for BasedOnStyle: ", BasedOnStyle)); 91 return; 92 } 93 } 94 95 IO.mapOptional("AccessModifierOffset", Style.AccessModifierOffset); 96 IO.mapOptional("ConstructorInitializerIndentWidth", 97 Style.ConstructorInitializerIndentWidth); 98 IO.mapOptional("AlignEscapedNewlinesLeft", Style.AlignEscapedNewlinesLeft); 99 IO.mapOptional("AlignTrailingComments", Style.AlignTrailingComments); 100 IO.mapOptional("AllowAllParametersOfDeclarationOnNextLine", 101 Style.AllowAllParametersOfDeclarationOnNextLine); 102 IO.mapOptional("AllowShortIfStatementsOnASingleLine", 103 Style.AllowShortIfStatementsOnASingleLine); 104 IO.mapOptional("AllowShortLoopsOnASingleLine", 105 Style.AllowShortLoopsOnASingleLine); 106 IO.mapOptional("AlwaysBreakTemplateDeclarations", 107 Style.AlwaysBreakTemplateDeclarations); 108 IO.mapOptional("AlwaysBreakBeforeMultilineStrings", 109 Style.AlwaysBreakBeforeMultilineStrings); 110 IO.mapOptional("BreakBeforeBinaryOperators", 111 Style.BreakBeforeBinaryOperators); 112 IO.mapOptional("BreakConstructorInitializersBeforeComma", 113 Style.BreakConstructorInitializersBeforeComma); 114 IO.mapOptional("BinPackParameters", Style.BinPackParameters); 115 IO.mapOptional("ColumnLimit", Style.ColumnLimit); 116 IO.mapOptional("ConstructorInitializerAllOnOneLineOrOnePerLine", 117 Style.ConstructorInitializerAllOnOneLineOrOnePerLine); 118 IO.mapOptional("DerivePointerBinding", Style.DerivePointerBinding); 119 IO.mapOptional("ExperimentalAutoDetectBinPacking", 120 Style.ExperimentalAutoDetectBinPacking); 121 IO.mapOptional("IndentCaseLabels", Style.IndentCaseLabels); 122 IO.mapOptional("MaxEmptyLinesToKeep", Style.MaxEmptyLinesToKeep); 123 IO.mapOptional("NamespaceIndentation", Style.NamespaceIndentation); 124 IO.mapOptional("ObjCSpaceBeforeProtocolList", 125 Style.ObjCSpaceBeforeProtocolList); 126 IO.mapOptional("PenaltyBreakComment", Style.PenaltyBreakComment); 127 IO.mapOptional("PenaltyBreakString", Style.PenaltyBreakString); 128 IO.mapOptional("PenaltyBreakFirstLessLess", 129 Style.PenaltyBreakFirstLessLess); 130 IO.mapOptional("PenaltyExcessCharacter", Style.PenaltyExcessCharacter); 131 IO.mapOptional("PenaltyReturnTypeOnItsOwnLine", 132 Style.PenaltyReturnTypeOnItsOwnLine); 133 IO.mapOptional("PointerBindsToType", Style.PointerBindsToType); 134 IO.mapOptional("SpacesBeforeTrailingComments", 135 Style.SpacesBeforeTrailingComments); 136 IO.mapOptional("Cpp11BracedListStyle", Style.Cpp11BracedListStyle); 137 IO.mapOptional("Standard", Style.Standard); 138 IO.mapOptional("IndentWidth", Style.IndentWidth); 139 IO.mapOptional("UseTab", Style.UseTab); 140 IO.mapOptional("BreakBeforeBraces", Style.BreakBeforeBraces); 141 IO.mapOptional("IndentFunctionDeclarationAfterType", 142 Style.IndentFunctionDeclarationAfterType); 143 IO.mapOptional("SpacesInParentheses", Style.SpacesInParentheses); 144 IO.mapOptional("SpaceInEmptyParentheses", Style.SpaceInEmptyParentheses); 145 IO.mapOptional("SpacesInCStyleCastParentheses", 146 Style.SpacesInCStyleCastParentheses); 147 IO.mapOptional("SpaceAfterControlStatementKeyword", 148 Style.SpaceAfterControlStatementKeyword); 149 } 150 }; 151 } 152 } 153 154 namespace clang { 155 namespace format { 156 157 void setDefaultPenalties(FormatStyle &Style) { 158 Style.PenaltyBreakComment = 60; 159 Style.PenaltyBreakFirstLessLess = 120; 160 Style.PenaltyBreakString = 1000; 161 Style.PenaltyExcessCharacter = 1000000; 162 } 163 164 FormatStyle getLLVMStyle() { 165 FormatStyle LLVMStyle; 166 LLVMStyle.AccessModifierOffset = -2; 167 LLVMStyle.AlignEscapedNewlinesLeft = false; 168 LLVMStyle.AlignTrailingComments = true; 169 LLVMStyle.AllowAllParametersOfDeclarationOnNextLine = true; 170 LLVMStyle.AllowShortIfStatementsOnASingleLine = false; 171 LLVMStyle.AllowShortLoopsOnASingleLine = false; 172 LLVMStyle.AlwaysBreakBeforeMultilineStrings = false; 173 LLVMStyle.AlwaysBreakTemplateDeclarations = false; 174 LLVMStyle.BinPackParameters = true; 175 LLVMStyle.BreakBeforeBinaryOperators = false; 176 LLVMStyle.BreakBeforeBraces = FormatStyle::BS_Attach; 177 LLVMStyle.BreakConstructorInitializersBeforeComma = false; 178 LLVMStyle.ColumnLimit = 80; 179 LLVMStyle.ConstructorInitializerAllOnOneLineOrOnePerLine = false; 180 LLVMStyle.ConstructorInitializerIndentWidth = 4; 181 LLVMStyle.Cpp11BracedListStyle = false; 182 LLVMStyle.DerivePointerBinding = false; 183 LLVMStyle.ExperimentalAutoDetectBinPacking = false; 184 LLVMStyle.IndentCaseLabels = false; 185 LLVMStyle.IndentFunctionDeclarationAfterType = false; 186 LLVMStyle.IndentWidth = 2; 187 LLVMStyle.MaxEmptyLinesToKeep = 1; 188 LLVMStyle.NamespaceIndentation = FormatStyle::NI_None; 189 LLVMStyle.ObjCSpaceBeforeProtocolList = true; 190 LLVMStyle.PointerBindsToType = false; 191 LLVMStyle.SpacesBeforeTrailingComments = 1; 192 LLVMStyle.Standard = FormatStyle::LS_Cpp03; 193 LLVMStyle.UseTab = false; 194 LLVMStyle.SpacesInParentheses = false; 195 LLVMStyle.SpaceInEmptyParentheses = false; 196 LLVMStyle.SpacesInCStyleCastParentheses = false; 197 LLVMStyle.SpaceAfterControlStatementKeyword = true; 198 199 setDefaultPenalties(LLVMStyle); 200 LLVMStyle.PenaltyReturnTypeOnItsOwnLine = 60; 201 202 return LLVMStyle; 203 } 204 205 FormatStyle getGoogleStyle() { 206 FormatStyle GoogleStyle; 207 GoogleStyle.AccessModifierOffset = -1; 208 GoogleStyle.AlignEscapedNewlinesLeft = true; 209 GoogleStyle.AlignTrailingComments = true; 210 GoogleStyle.AllowAllParametersOfDeclarationOnNextLine = true; 211 GoogleStyle.AllowShortIfStatementsOnASingleLine = true; 212 GoogleStyle.AllowShortLoopsOnASingleLine = true; 213 GoogleStyle.AlwaysBreakBeforeMultilineStrings = true; 214 GoogleStyle.AlwaysBreakTemplateDeclarations = true; 215 GoogleStyle.BinPackParameters = true; 216 GoogleStyle.BreakBeforeBinaryOperators = false; 217 GoogleStyle.BreakBeforeBraces = FormatStyle::BS_Attach; 218 GoogleStyle.BreakConstructorInitializersBeforeComma = false; 219 GoogleStyle.ColumnLimit = 80; 220 GoogleStyle.ConstructorInitializerAllOnOneLineOrOnePerLine = true; 221 GoogleStyle.ConstructorInitializerIndentWidth = 4; 222 GoogleStyle.Cpp11BracedListStyle = true; 223 GoogleStyle.DerivePointerBinding = true; 224 GoogleStyle.ExperimentalAutoDetectBinPacking = false; 225 GoogleStyle.IndentCaseLabels = true; 226 GoogleStyle.IndentFunctionDeclarationAfterType = true; 227 GoogleStyle.IndentWidth = 2; 228 GoogleStyle.MaxEmptyLinesToKeep = 1; 229 GoogleStyle.NamespaceIndentation = FormatStyle::NI_None; 230 GoogleStyle.ObjCSpaceBeforeProtocolList = false; 231 GoogleStyle.PointerBindsToType = true; 232 GoogleStyle.SpacesBeforeTrailingComments = 2; 233 GoogleStyle.Standard = FormatStyle::LS_Auto; 234 GoogleStyle.UseTab = false; 235 GoogleStyle.SpacesInParentheses = false; 236 GoogleStyle.SpaceInEmptyParentheses = false; 237 GoogleStyle.SpacesInCStyleCastParentheses = false; 238 GoogleStyle.SpaceAfterControlStatementKeyword = true; 239 240 setDefaultPenalties(GoogleStyle); 241 GoogleStyle.PenaltyReturnTypeOnItsOwnLine = 200; 242 243 return GoogleStyle; 244 } 245 246 FormatStyle getChromiumStyle() { 247 FormatStyle ChromiumStyle = getGoogleStyle(); 248 ChromiumStyle.AllowAllParametersOfDeclarationOnNextLine = false; 249 ChromiumStyle.AllowShortIfStatementsOnASingleLine = false; 250 ChromiumStyle.AllowShortLoopsOnASingleLine = false; 251 ChromiumStyle.BinPackParameters = false; 252 ChromiumStyle.DerivePointerBinding = false; 253 ChromiumStyle.Standard = FormatStyle::LS_Cpp03; 254 return ChromiumStyle; 255 } 256 257 FormatStyle getMozillaStyle() { 258 FormatStyle MozillaStyle = getLLVMStyle(); 259 MozillaStyle.AllowAllParametersOfDeclarationOnNextLine = false; 260 MozillaStyle.ConstructorInitializerAllOnOneLineOrOnePerLine = true; 261 MozillaStyle.DerivePointerBinding = true; 262 MozillaStyle.IndentCaseLabels = true; 263 MozillaStyle.ObjCSpaceBeforeProtocolList = false; 264 MozillaStyle.PenaltyReturnTypeOnItsOwnLine = 200; 265 MozillaStyle.PointerBindsToType = true; 266 return MozillaStyle; 267 } 268 269 FormatStyle getWebKitStyle() { 270 FormatStyle Style = getLLVMStyle(); 271 Style.AccessModifierOffset = -4; 272 Style.AlignTrailingComments = false; 273 Style.BreakBeforeBinaryOperators = true; 274 Style.BreakBeforeBraces = FormatStyle::BS_Stroustrup; 275 Style.BreakConstructorInitializersBeforeComma = true; 276 Style.ColumnLimit = 0; 277 Style.IndentWidth = 4; 278 Style.NamespaceIndentation = FormatStyle::NI_Inner; 279 Style.PointerBindsToType = true; 280 return Style; 281 } 282 283 bool getPredefinedStyle(StringRef Name, FormatStyle *Style) { 284 if (Name.equals_lower("llvm")) 285 *Style = getLLVMStyle(); 286 else if (Name.equals_lower("chromium")) 287 *Style = getChromiumStyle(); 288 else if (Name.equals_lower("mozilla")) 289 *Style = getMozillaStyle(); 290 else if (Name.equals_lower("google")) 291 *Style = getGoogleStyle(); 292 else if (Name.equals_lower("webkit")) 293 *Style = getWebKitStyle(); 294 else 295 return false; 296 297 return true; 298 } 299 300 llvm::error_code parseConfiguration(StringRef Text, FormatStyle *Style) { 301 if (Text.trim().empty()) 302 return llvm::make_error_code(llvm::errc::invalid_argument); 303 llvm::yaml::Input Input(Text); 304 Input >> *Style; 305 return Input.error(); 306 } 307 308 std::string configurationAsText(const FormatStyle &Style) { 309 std::string Text; 310 llvm::raw_string_ostream Stream(Text); 311 llvm::yaml::Output Output(Stream); 312 // We use the same mapping method for input and output, so we need a non-const 313 // reference here. 314 FormatStyle NonConstStyle = Style; 315 Output << NonConstStyle; 316 return Stream.str(); 317 } 318 319 namespace { 320 321 class NoColumnLimitFormatter { 322 public: 323 NoColumnLimitFormatter(ContinuationIndenter *Indenter) : Indenter(Indenter) {} 324 325 /// \brief Formats the line starting at \p State, simply keeping all of the 326 /// input's line breaking decisions. 327 void format() { 328 LineState State = Indenter->getInitialState(); 329 while (State.NextToken != NULL) { 330 bool Newline = 331 Indenter->mustBreak(State) || 332 (Indenter->canBreak(State) && State.NextToken->NewlinesBefore > 0); 333 Indenter->addTokenToState(State, Newline, /*DryRun=*/false); 334 } 335 } 336 337 private: 338 ContinuationIndenter *Indenter; 339 }; 340 341 class UnwrappedLineFormatter { 342 public: 343 UnwrappedLineFormatter(ContinuationIndenter *Indenter, 344 const FormatStyle &Style, const AnnotatedLine &Line) 345 : Indenter(Indenter), Style(Style), Line(Line), Count(0) {} 346 347 /// \brief Formats an \c UnwrappedLine. 348 void format() { 349 LineState State = Indenter->getInitialState(); 350 351 // If the ObjC method declaration does not fit on a line, we should format 352 // it with one arg per line. 353 if (Line.Type == LT_ObjCMethodDecl) 354 State.Stack.back().BreakBeforeParameter = true; 355 356 // Find best solution in solution space. 357 analyzeSolutionSpace(State); 358 } 359 360 private: 361 /// \brief An edge in the solution space from \c Previous->State to \c State, 362 /// inserting a newline dependent on the \c NewLine. 363 struct StateNode { 364 StateNode(const LineState &State, bool NewLine, StateNode *Previous) 365 : State(State), NewLine(NewLine), Previous(Previous) {} 366 LineState State; 367 bool NewLine; 368 StateNode *Previous; 369 }; 370 371 /// \brief A pair of <penalty, count> that is used to prioritize the BFS on. 372 /// 373 /// In case of equal penalties, we want to prefer states that were inserted 374 /// first. During state generation we make sure that we insert states first 375 /// that break the line as late as possible. 376 typedef std::pair<unsigned, unsigned> OrderedPenalty; 377 378 /// \brief An item in the prioritized BFS search queue. The \c StateNode's 379 /// \c State has the given \c OrderedPenalty. 380 typedef std::pair<OrderedPenalty, StateNode *> QueueItem; 381 382 /// \brief The BFS queue type. 383 typedef std::priority_queue<QueueItem, std::vector<QueueItem>, 384 std::greater<QueueItem> > QueueType; 385 386 /// \brief Analyze the entire solution space starting from \p InitialState. 387 /// 388 /// This implements a variant of Dijkstra's algorithm on the graph that spans 389 /// the solution space (\c LineStates are the nodes). The algorithm tries to 390 /// find the shortest path (the one with lowest penalty) from \p InitialState 391 /// to a state where all tokens are placed. 392 void analyzeSolutionSpace(LineState &InitialState) { 393 std::set<LineState> Seen; 394 395 // Insert start element into queue. 396 StateNode *Node = 397 new (Allocator.Allocate()) StateNode(InitialState, false, NULL); 398 Queue.push(QueueItem(OrderedPenalty(0, Count), Node)); 399 ++Count; 400 401 // While not empty, take first element and follow edges. 402 while (!Queue.empty()) { 403 unsigned Penalty = Queue.top().first.first; 404 StateNode *Node = Queue.top().second; 405 if (Node->State.NextToken == NULL) { 406 DEBUG(llvm::dbgs() << "\n---\nPenalty for line: " << Penalty << "\n"); 407 break; 408 } 409 Queue.pop(); 410 411 // Cut off the analysis of certain solutions if the analysis gets too 412 // complex. See description of IgnoreStackForComparison. 413 if (Count > 10000) 414 Node->State.IgnoreStackForComparison = true; 415 416 if (!Seen.insert(Node->State).second) 417 // State already examined with lower penalty. 418 continue; 419 420 addNextStateToQueue(Penalty, Node, /*NewLine=*/false); 421 addNextStateToQueue(Penalty, Node, /*NewLine=*/true); 422 } 423 424 if (Queue.empty()) 425 // We were unable to find a solution, do nothing. 426 // FIXME: Add diagnostic? 427 return; 428 429 // Reconstruct the solution. 430 reconstructPath(InitialState, Queue.top().second); 431 DEBUG(llvm::dbgs() << "Total number of analyzed states: " << Count << "\n"); 432 DEBUG(llvm::dbgs() << "---\n"); 433 } 434 435 void reconstructPath(LineState &State, StateNode *Current) { 436 std::deque<StateNode *> Path; 437 // We do not need a break before the initial token. 438 while (Current->Previous) { 439 Path.push_front(Current); 440 Current = Current->Previous; 441 } 442 for (std::deque<StateNode *>::iterator I = Path.begin(), E = Path.end(); 443 I != E; ++I) { 444 unsigned Penalty = Indenter->addTokenToState(State, (*I)->NewLine, false); 445 (void)Penalty; 446 DEBUG({ 447 if ((*I)->NewLine) { 448 llvm::dbgs() << "Penalty for placing " 449 << (*I)->Previous->State.NextToken->Tok.getName() << ": " 450 << Penalty << "\n"; 451 } 452 }); 453 } 454 } 455 456 /// \brief Add the following state to the analysis queue \c Queue. 457 /// 458 /// Assume the current state is \p PreviousNode and has been reached with a 459 /// penalty of \p Penalty. Insert a line break if \p NewLine is \c true. 460 void addNextStateToQueue(unsigned Penalty, StateNode *PreviousNode, 461 bool NewLine) { 462 if (NewLine && !Indenter->canBreak(PreviousNode->State)) 463 return; 464 if (!NewLine && Indenter->mustBreak(PreviousNode->State)) 465 return; 466 467 StateNode *Node = new (Allocator.Allocate()) 468 StateNode(PreviousNode->State, NewLine, PreviousNode); 469 Penalty += Indenter->addTokenToState(Node->State, NewLine, true); 470 if (Node->State.Column > Indenter->getColumnLimit()) { 471 unsigned ExcessCharacters = 472 Node->State.Column - Indenter->getColumnLimit(); 473 Penalty += Style.PenaltyExcessCharacter * ExcessCharacters; 474 } 475 476 Queue.push(QueueItem(OrderedPenalty(Penalty, Count), Node)); 477 ++Count; 478 } 479 480 ContinuationIndenter *Indenter; 481 FormatStyle Style; 482 const AnnotatedLine &Line; 483 484 llvm::SpecificBumpPtrAllocator<StateNode> Allocator; 485 QueueType Queue; 486 // Increasing count of \c StateNode items we have created. This is used 487 // to create a deterministic order independent of the container. 488 unsigned Count; 489 }; 490 491 class FormatTokenLexer { 492 public: 493 FormatTokenLexer(Lexer &Lex, SourceManager &SourceMgr, FormatStyle &Style, 494 encoding::Encoding Encoding) 495 : FormatTok(NULL), GreaterStashed(false), Column(0), 496 TrailingWhitespace(0), Lex(Lex), SourceMgr(SourceMgr), Style(Style), 497 IdentTable(getFormattingLangOpts()), Encoding(Encoding) { 498 Lex.SetKeepWhitespaceMode(true); 499 } 500 501 ArrayRef<FormatToken *> lex() { 502 assert(Tokens.empty()); 503 do { 504 Tokens.push_back(getNextToken()); 505 } while (Tokens.back()->Tok.isNot(tok::eof)); 506 return Tokens; 507 } 508 509 IdentifierTable &getIdentTable() { return IdentTable; } 510 511 private: 512 FormatToken *getNextToken() { 513 if (GreaterStashed) { 514 // Create a synthesized second '>' token. 515 // FIXME: Increment Column and set OriginalColumn. 516 Token Greater = FormatTok->Tok; 517 FormatTok = new (Allocator.Allocate()) FormatToken; 518 FormatTok->Tok = Greater; 519 SourceLocation GreaterLocation = 520 FormatTok->Tok.getLocation().getLocWithOffset(1); 521 FormatTok->WhitespaceRange = 522 SourceRange(GreaterLocation, GreaterLocation); 523 FormatTok->TokenText = ">"; 524 FormatTok->CodePointCount = 1; 525 GreaterStashed = false; 526 return FormatTok; 527 } 528 529 FormatTok = new (Allocator.Allocate()) FormatToken; 530 readRawToken(*FormatTok); 531 SourceLocation WhitespaceStart = 532 FormatTok->Tok.getLocation().getLocWithOffset(-TrailingWhitespace); 533 if (SourceMgr.getFileOffset(WhitespaceStart) == 0) 534 FormatTok->IsFirst = true; 535 536 // Consume and record whitespace until we find a significant token. 537 unsigned WhitespaceLength = TrailingWhitespace; 538 while (FormatTok->Tok.is(tok::unknown)) { 539 for (int i = 0, e = FormatTok->TokenText.size(); i != e; ++i) { 540 switch (FormatTok->TokenText[i]) { 541 case '\n': 542 ++FormatTok->NewlinesBefore; 543 // FIXME: This is technically incorrect, as it could also 544 // be a literal backslash at the end of the line. 545 if (i == 0 || FormatTok->TokenText[i-1] != '\\') 546 FormatTok->HasUnescapedNewline = true; 547 FormatTok->LastNewlineOffset = WhitespaceLength + i + 1; 548 Column = 0; 549 break; 550 case ' ': 551 ++Column; 552 break; 553 case '\t': 554 Column += Style.IndentWidth - Column % Style.IndentWidth; 555 break; 556 default: 557 ++Column; 558 break; 559 } 560 } 561 562 WhitespaceLength += FormatTok->Tok.getLength(); 563 564 readRawToken(*FormatTok); 565 } 566 567 // In case the token starts with escaped newlines, we want to 568 // take them into account as whitespace - this pattern is quite frequent 569 // in macro definitions. 570 // FIXME: What do we want to do with other escaped spaces, and escaped 571 // spaces or newlines in the middle of tokens? 572 // FIXME: Add a more explicit test. 573 while (FormatTok->TokenText.size() > 1 && FormatTok->TokenText[0] == '\\' && 574 FormatTok->TokenText[1] == '\n') { 575 // FIXME: ++FormatTok->NewlinesBefore is missing... 576 WhitespaceLength += 2; 577 Column = 0; 578 FormatTok->TokenText = FormatTok->TokenText.substr(2); 579 } 580 FormatTok->OriginalColumn = Column; 581 582 TrailingWhitespace = 0; 583 if (FormatTok->Tok.is(tok::comment)) { 584 // FIXME: Add the trimmed whitespace to Column. 585 StringRef UntrimmedText = FormatTok->TokenText; 586 FormatTok->TokenText = FormatTok->TokenText.rtrim(); 587 TrailingWhitespace = UntrimmedText.size() - FormatTok->TokenText.size(); 588 } else if (FormatTok->Tok.is(tok::raw_identifier)) { 589 IdentifierInfo &Info = IdentTable.get(FormatTok->TokenText); 590 FormatTok->Tok.setIdentifierInfo(&Info); 591 FormatTok->Tok.setKind(Info.getTokenID()); 592 } else if (FormatTok->Tok.is(tok::greatergreater)) { 593 FormatTok->Tok.setKind(tok::greater); 594 FormatTok->TokenText = FormatTok->TokenText.substr(0, 1); 595 GreaterStashed = true; 596 } 597 598 // Now FormatTok is the next non-whitespace token. 599 FormatTok->CodePointCount = 600 encoding::getCodePointCount(FormatTok->TokenText, Encoding); 601 602 if (FormatTok->isOneOf(tok::string_literal, tok::comment)) { 603 StringRef Text = FormatTok->TokenText; 604 size_t FirstNewlinePos = Text.find('\n'); 605 if (FirstNewlinePos != StringRef::npos) { 606 FormatTok->CodePointsInFirstLine = encoding::getCodePointCount( 607 Text.substr(0, FirstNewlinePos), Encoding); 608 FormatTok->CodePointsInLastLine = encoding::getCodePointCount( 609 Text.substr(Text.find_last_of('\n') + 1), Encoding); 610 } 611 } 612 // FIXME: Add the CodePointCount to Column. 613 FormatTok->WhitespaceRange = SourceRange( 614 WhitespaceStart, WhitespaceStart.getLocWithOffset(WhitespaceLength)); 615 return FormatTok; 616 } 617 618 FormatToken *FormatTok; 619 bool GreaterStashed; 620 unsigned Column; 621 unsigned TrailingWhitespace; 622 Lexer &Lex; 623 SourceManager &SourceMgr; 624 FormatStyle &Style; 625 IdentifierTable IdentTable; 626 encoding::Encoding Encoding; 627 llvm::SpecificBumpPtrAllocator<FormatToken> Allocator; 628 SmallVector<FormatToken *, 16> Tokens; 629 630 void readRawToken(FormatToken &Tok) { 631 Lex.LexFromRawLexer(Tok.Tok); 632 Tok.TokenText = StringRef(SourceMgr.getCharacterData(Tok.Tok.getLocation()), 633 Tok.Tok.getLength()); 634 // For formatting, treat unterminated string literals like normal string 635 // literals. 636 if (Tok.is(tok::unknown) && !Tok.TokenText.empty() && 637 Tok.TokenText[0] == '"') { 638 Tok.Tok.setKind(tok::string_literal); 639 Tok.IsUnterminatedLiteral = true; 640 } 641 } 642 }; 643 644 class Formatter : public UnwrappedLineConsumer { 645 public: 646 Formatter(const FormatStyle &Style, Lexer &Lex, SourceManager &SourceMgr, 647 const std::vector<CharSourceRange> &Ranges) 648 : Style(Style), Lex(Lex), SourceMgr(SourceMgr), 649 Whitespaces(SourceMgr, Style), Ranges(Ranges), 650 Encoding(encoding::detectEncoding(Lex.getBuffer())) { 651 DEBUG(llvm::dbgs() << "File encoding: " 652 << (Encoding == encoding::Encoding_UTF8 ? "UTF8" 653 : "unknown") 654 << "\n"); 655 } 656 657 virtual ~Formatter() {} 658 659 tooling::Replacements format() { 660 FormatTokenLexer Tokens(Lex, SourceMgr, Style, Encoding); 661 662 UnwrappedLineParser Parser(Style, Tokens.lex(), *this); 663 bool StructuralError = Parser.parse(); 664 TokenAnnotator Annotator(Style, Tokens.getIdentTable().get("in")); 665 for (unsigned i = 0, e = AnnotatedLines.size(); i != e; ++i) { 666 Annotator.annotate(AnnotatedLines[i]); 667 } 668 deriveLocalStyle(); 669 for (unsigned i = 0, e = AnnotatedLines.size(); i != e; ++i) { 670 Annotator.calculateFormattingInformation(AnnotatedLines[i]); 671 } 672 673 // Adapt level to the next line if this is a comment. 674 // FIXME: Can/should this be done in the UnwrappedLineParser? 675 const AnnotatedLine *NextNonCommentLine = NULL; 676 for (unsigned i = AnnotatedLines.size() - 1; i > 0; --i) { 677 if (NextNonCommentLine && AnnotatedLines[i].First->is(tok::comment) && 678 !AnnotatedLines[i].First->Next) 679 AnnotatedLines[i].Level = NextNonCommentLine->Level; 680 else 681 NextNonCommentLine = AnnotatedLines[i].First->isNot(tok::r_brace) 682 ? &AnnotatedLines[i] 683 : NULL; 684 } 685 686 std::vector<int> IndentForLevel; 687 bool PreviousLineWasTouched = false; 688 const FormatToken *PreviousLineLastToken = 0; 689 bool FormatPPDirective = false; 690 for (std::vector<AnnotatedLine>::iterator I = AnnotatedLines.begin(), 691 E = AnnotatedLines.end(); 692 I != E; ++I) { 693 const AnnotatedLine &TheLine = *I; 694 const FormatToken *FirstTok = TheLine.First; 695 int Offset = getIndentOffset(*TheLine.First); 696 697 // Check whether this line is part of a formatted preprocessor directive. 698 if (FirstTok->HasUnescapedNewline) 699 FormatPPDirective = false; 700 if (!FormatPPDirective && TheLine.InPPDirective && 701 (touchesLine(TheLine) || touchesPPDirective(I + 1, E))) 702 FormatPPDirective = true; 703 704 // Determine indent and try to merge multiple unwrapped lines. 705 while (IndentForLevel.size() <= TheLine.Level) 706 IndentForLevel.push_back(-1); 707 IndentForLevel.resize(TheLine.Level + 1); 708 unsigned Indent = getIndent(IndentForLevel, TheLine.Level); 709 if (static_cast<int>(Indent) + Offset >= 0) 710 Indent += Offset; 711 tryFitMultipleLinesInOne(Indent, I, E); 712 713 bool WasMoved = PreviousLineWasTouched && FirstTok->NewlinesBefore == 0; 714 if (TheLine.First->is(tok::eof)) { 715 if (PreviousLineWasTouched) { 716 unsigned NewLines = std::min(FirstTok->NewlinesBefore, 1u); 717 Whitespaces.replaceWhitespace(*TheLine.First, NewLines, /*Indent*/ 0, 718 /*TargetColumn*/ 0); 719 } 720 } else if (TheLine.Type != LT_Invalid && 721 (WasMoved || FormatPPDirective || touchesLine(TheLine))) { 722 unsigned LevelIndent = getIndent(IndentForLevel, TheLine.Level); 723 if (FirstTok->WhitespaceRange.isValid() && 724 // Insert a break even if there is a structural error in case where 725 // we break apart a line consisting of multiple unwrapped lines. 726 (FirstTok->NewlinesBefore == 0 || !StructuralError)) { 727 formatFirstToken(*TheLine.First, PreviousLineLastToken, Indent, 728 TheLine.InPPDirective); 729 } else { 730 Indent = LevelIndent = FirstTok->OriginalColumn; 731 } 732 ContinuationIndenter Indenter(Style, SourceMgr, TheLine, Indent, 733 Whitespaces, Encoding, 734 BinPackInconclusiveFunctions); 735 736 // If everything fits on a single line, just put it there. 737 unsigned ColumnLimit = Style.ColumnLimit; 738 if ((I + 1) != E && (I + 1)->InPPDirective && 739 !(I + 1)->First->HasUnescapedNewline) 740 ColumnLimit = Indenter.getColumnLimit(); 741 742 if (I->Last->TotalLength + Indent <= ColumnLimit) { 743 LineState State = Indenter.getInitialState(); 744 while (State.NextToken != NULL) 745 Indenter.addTokenToState(State, false, false); 746 } else if (Style.ColumnLimit == 0) { 747 NoColumnLimitFormatter Formatter(&Indenter); 748 Formatter.format(); 749 } else { 750 UnwrappedLineFormatter Formatter(&Indenter, Style, TheLine); 751 Formatter.format(); 752 } 753 754 IndentForLevel[TheLine.Level] = LevelIndent; 755 PreviousLineWasTouched = true; 756 } else { 757 // Format the first token if necessary, and notify the WhitespaceManager 758 // about the unchanged whitespace. 759 for (const FormatToken *Tok = TheLine.First; Tok != NULL; 760 Tok = Tok->Next) { 761 if (Tok == TheLine.First && 762 (Tok->NewlinesBefore > 0 || Tok->IsFirst)) { 763 unsigned LevelIndent = Tok->OriginalColumn; 764 // Remove trailing whitespace of the previous line if it was 765 // touched. 766 if (PreviousLineWasTouched || touchesEmptyLineBefore(TheLine)) { 767 formatFirstToken(*Tok, PreviousLineLastToken, LevelIndent, 768 TheLine.InPPDirective); 769 } else { 770 Whitespaces.addUntouchableToken(*Tok, TheLine.InPPDirective); 771 } 772 773 if (static_cast<int>(LevelIndent) - Offset >= 0) 774 LevelIndent -= Offset; 775 if (Tok->isNot(tok::comment)) 776 IndentForLevel[TheLine.Level] = LevelIndent; 777 } else { 778 Whitespaces.addUntouchableToken(*Tok, TheLine.InPPDirective); 779 } 780 } 781 // If we did not reformat this unwrapped line, the column at the end of 782 // the last token is unchanged - thus, we can calculate the end of the 783 // last token. 784 PreviousLineWasTouched = false; 785 } 786 PreviousLineLastToken = I->Last; 787 } 788 return Whitespaces.generateReplacements(); 789 } 790 791 private: 792 void deriveLocalStyle() { 793 unsigned CountBoundToVariable = 0; 794 unsigned CountBoundToType = 0; 795 bool HasCpp03IncompatibleFormat = false; 796 bool HasBinPackedFunction = false; 797 bool HasOnePerLineFunction = false; 798 for (unsigned i = 0, e = AnnotatedLines.size(); i != e; ++i) { 799 if (!AnnotatedLines[i].First->Next) 800 continue; 801 FormatToken *Tok = AnnotatedLines[i].First->Next; 802 while (Tok->Next) { 803 if (Tok->Type == TT_PointerOrReference) { 804 bool SpacesBefore = 805 Tok->WhitespaceRange.getBegin() != Tok->WhitespaceRange.getEnd(); 806 bool SpacesAfter = Tok->Next->WhitespaceRange.getBegin() != 807 Tok->Next->WhitespaceRange.getEnd(); 808 if (SpacesBefore && !SpacesAfter) 809 ++CountBoundToVariable; 810 else if (!SpacesBefore && SpacesAfter) 811 ++CountBoundToType; 812 } 813 814 if (Tok->Type == TT_TemplateCloser && 815 Tok->Previous->Type == TT_TemplateCloser && 816 Tok->WhitespaceRange.getBegin() == Tok->WhitespaceRange.getEnd()) 817 HasCpp03IncompatibleFormat = true; 818 819 if (Tok->PackingKind == PPK_BinPacked) 820 HasBinPackedFunction = true; 821 if (Tok->PackingKind == PPK_OnePerLine) 822 HasOnePerLineFunction = true; 823 824 Tok = Tok->Next; 825 } 826 } 827 if (Style.DerivePointerBinding) { 828 if (CountBoundToType > CountBoundToVariable) 829 Style.PointerBindsToType = true; 830 else if (CountBoundToType < CountBoundToVariable) 831 Style.PointerBindsToType = false; 832 } 833 if (Style.Standard == FormatStyle::LS_Auto) { 834 Style.Standard = HasCpp03IncompatibleFormat ? FormatStyle::LS_Cpp11 835 : FormatStyle::LS_Cpp03; 836 } 837 BinPackInconclusiveFunctions = 838 HasBinPackedFunction || !HasOnePerLineFunction; 839 } 840 841 /// \brief Get the indent of \p Level from \p IndentForLevel. 842 /// 843 /// \p IndentForLevel must contain the indent for the level \c l 844 /// at \p IndentForLevel[l], or a value < 0 if the indent for 845 /// that level is unknown. 846 unsigned getIndent(const std::vector<int> IndentForLevel, unsigned Level) { 847 if (IndentForLevel[Level] != -1) 848 return IndentForLevel[Level]; 849 if (Level == 0) 850 return 0; 851 return getIndent(IndentForLevel, Level - 1) + Style.IndentWidth; 852 } 853 854 /// \brief Get the offset of the line relatively to the level. 855 /// 856 /// For example, 'public:' labels in classes are offset by 1 or 2 857 /// characters to the left from their level. 858 int getIndentOffset(const FormatToken &RootToken) { 859 if (RootToken.isAccessSpecifier(false) || RootToken.isObjCAccessSpecifier()) 860 return Style.AccessModifierOffset; 861 return 0; 862 } 863 864 /// \brief Tries to merge lines into one. 865 /// 866 /// This will change \c Line and \c AnnotatedLine to contain the merged line, 867 /// if possible; note that \c I will be incremented when lines are merged. 868 void tryFitMultipleLinesInOne(unsigned Indent, 869 std::vector<AnnotatedLine>::iterator &I, 870 std::vector<AnnotatedLine>::iterator E) { 871 // We can never merge stuff if there are trailing line comments. 872 if (I->Last->Type == TT_LineComment) 873 return; 874 875 if (Indent > Style.ColumnLimit) 876 return; 877 878 unsigned Limit = Style.ColumnLimit - Indent; 879 // If we already exceed the column limit, we set 'Limit' to 0. The different 880 // tryMerge..() functions can then decide whether to still do merging. 881 Limit = I->Last->TotalLength > Limit ? 0 : Limit - I->Last->TotalLength; 882 883 if (I + 1 == E || (I + 1)->Type == LT_Invalid) 884 return; 885 886 if (I->Last->is(tok::l_brace)) { 887 tryMergeSimpleBlock(I, E, Limit); 888 } else if (Style.AllowShortIfStatementsOnASingleLine && 889 I->First->is(tok::kw_if)) { 890 tryMergeSimpleControlStatement(I, E, Limit); 891 } else if (Style.AllowShortLoopsOnASingleLine && 892 I->First->isOneOf(tok::kw_for, tok::kw_while)) { 893 tryMergeSimpleControlStatement(I, E, Limit); 894 } else if (I->InPPDirective && 895 (I->First->HasUnescapedNewline || I->First->IsFirst)) { 896 tryMergeSimplePPDirective(I, E, Limit); 897 } 898 } 899 900 void tryMergeSimplePPDirective(std::vector<AnnotatedLine>::iterator &I, 901 std::vector<AnnotatedLine>::iterator E, 902 unsigned Limit) { 903 if (Limit == 0) 904 return; 905 AnnotatedLine &Line = *I; 906 if (!(I + 1)->InPPDirective || (I + 1)->First->HasUnescapedNewline) 907 return; 908 if (I + 2 != E && (I + 2)->InPPDirective && 909 !(I + 2)->First->HasUnescapedNewline) 910 return; 911 if (1 + (I + 1)->Last->TotalLength > Limit) 912 return; 913 join(Line, *(++I)); 914 } 915 916 void tryMergeSimpleControlStatement(std::vector<AnnotatedLine>::iterator &I, 917 std::vector<AnnotatedLine>::iterator E, 918 unsigned Limit) { 919 if (Limit == 0) 920 return; 921 if (Style.BreakBeforeBraces == FormatStyle::BS_Allman && 922 (I + 1)->First->is(tok::l_brace)) 923 return; 924 if ((I + 1)->InPPDirective != I->InPPDirective || 925 ((I + 1)->InPPDirective && (I + 1)->First->HasUnescapedNewline)) 926 return; 927 AnnotatedLine &Line = *I; 928 if (Line.Last->isNot(tok::r_paren)) 929 return; 930 if (1 + (I + 1)->Last->TotalLength > Limit) 931 return; 932 if ((I + 1)->First->isOneOf(tok::semi, tok::kw_if, tok::kw_for, 933 tok::kw_while) || 934 (I + 1)->First->Type == TT_LineComment) 935 return; 936 // Only inline simple if's (no nested if or else). 937 if (I + 2 != E && Line.First->is(tok::kw_if) && 938 (I + 2)->First->is(tok::kw_else)) 939 return; 940 join(Line, *(++I)); 941 } 942 943 void tryMergeSimpleBlock(std::vector<AnnotatedLine>::iterator &I, 944 std::vector<AnnotatedLine>::iterator E, 945 unsigned Limit) { 946 // No merging if the brace already is on the next line. 947 if (Style.BreakBeforeBraces != FormatStyle::BS_Attach) 948 return; 949 950 // First, check that the current line allows merging. This is the case if 951 // we're not in a control flow statement and the last token is an opening 952 // brace. 953 AnnotatedLine &Line = *I; 954 if (Line.First->isOneOf(tok::kw_if, tok::kw_while, tok::kw_do, tok::r_brace, 955 tok::kw_else, tok::kw_try, tok::kw_catch, 956 tok::kw_for, 957 // This gets rid of all ObjC @ keywords and methods. 958 tok::at, tok::minus, tok::plus)) 959 return; 960 961 FormatToken *Tok = (I + 1)->First; 962 if (Tok->is(tok::r_brace) && !Tok->MustBreakBefore && 963 (Tok->getNextNonComment() == NULL || 964 Tok->getNextNonComment()->is(tok::semi))) { 965 // We merge empty blocks even if the line exceeds the column limit. 966 Tok->SpacesRequiredBefore = 0; 967 Tok->CanBreakBefore = true; 968 join(Line, *(I + 1)); 969 I += 1; 970 } else if (Limit != 0 && Line.First->isNot(tok::kw_namespace)) { 971 // Check that we still have three lines and they fit into the limit. 972 if (I + 2 == E || (I + 2)->Type == LT_Invalid || 973 !nextTwoLinesFitInto(I, Limit)) 974 return; 975 976 // Second, check that the next line does not contain any braces - if it 977 // does, readability declines when putting it into a single line. 978 if ((I + 1)->Last->Type == TT_LineComment || Tok->MustBreakBefore) 979 return; 980 do { 981 if (Tok->isOneOf(tok::l_brace, tok::r_brace)) 982 return; 983 Tok = Tok->Next; 984 } while (Tok != NULL); 985 986 // Last, check that the third line contains a single closing brace. 987 Tok = (I + 2)->First; 988 if (Tok->getNextNonComment() != NULL || Tok->isNot(tok::r_brace) || 989 Tok->MustBreakBefore) 990 return; 991 992 join(Line, *(I + 1)); 993 join(Line, *(I + 2)); 994 I += 2; 995 } 996 } 997 998 bool nextTwoLinesFitInto(std::vector<AnnotatedLine>::iterator I, 999 unsigned Limit) { 1000 return 1 + (I + 1)->Last->TotalLength + 1 + (I + 2)->Last->TotalLength <= 1001 Limit; 1002 } 1003 1004 void join(AnnotatedLine &A, const AnnotatedLine &B) { 1005 assert(!A.Last->Next); 1006 assert(!B.First->Previous); 1007 A.Last->Next = B.First; 1008 B.First->Previous = A.Last; 1009 unsigned LengthA = A.Last->TotalLength + B.First->SpacesRequiredBefore; 1010 for (FormatToken *Tok = B.First; Tok; Tok = Tok->Next) { 1011 Tok->TotalLength += LengthA; 1012 A.Last = Tok; 1013 } 1014 } 1015 1016 bool touchesRanges(const CharSourceRange &Range) { 1017 for (unsigned i = 0, e = Ranges.size(); i != e; ++i) { 1018 if (!SourceMgr.isBeforeInTranslationUnit(Range.getEnd(), 1019 Ranges[i].getBegin()) && 1020 !SourceMgr.isBeforeInTranslationUnit(Ranges[i].getEnd(), 1021 Range.getBegin())) 1022 return true; 1023 } 1024 return false; 1025 } 1026 1027 bool touchesLine(const AnnotatedLine &TheLine) { 1028 const FormatToken *First = TheLine.First; 1029 const FormatToken *Last = TheLine.Last; 1030 CharSourceRange LineRange = CharSourceRange::getCharRange( 1031 First->WhitespaceRange.getBegin().getLocWithOffset( 1032 First->LastNewlineOffset), 1033 Last->Tok.getLocation().getLocWithOffset(Last->TokenText.size() - 1)); 1034 return touchesRanges(LineRange); 1035 } 1036 1037 bool touchesPPDirective(std::vector<AnnotatedLine>::iterator I, 1038 std::vector<AnnotatedLine>::iterator E) { 1039 for (; I != E; ++I) { 1040 if (I->First->HasUnescapedNewline) 1041 return false; 1042 if (touchesLine(*I)) 1043 return true; 1044 } 1045 return false; 1046 } 1047 1048 bool touchesEmptyLineBefore(const AnnotatedLine &TheLine) { 1049 const FormatToken *First = TheLine.First; 1050 CharSourceRange LineRange = CharSourceRange::getCharRange( 1051 First->WhitespaceRange.getBegin(), 1052 First->WhitespaceRange.getBegin().getLocWithOffset( 1053 First->LastNewlineOffset)); 1054 return touchesRanges(LineRange); 1055 } 1056 1057 virtual void consumeUnwrappedLine(const UnwrappedLine &TheLine) { 1058 AnnotatedLines.push_back(AnnotatedLine(TheLine)); 1059 } 1060 1061 /// \brief Add a new line and the required indent before the first Token 1062 /// of the \c UnwrappedLine if there was no structural parsing error. 1063 /// Returns the indent level of the \c UnwrappedLine. 1064 void formatFirstToken(const FormatToken &RootToken, 1065 const FormatToken *PreviousToken, unsigned Indent, 1066 bool InPPDirective) { 1067 unsigned Newlines = 1068 std::min(RootToken.NewlinesBefore, Style.MaxEmptyLinesToKeep + 1); 1069 // Remove empty lines before "}" where applicable. 1070 if (RootToken.is(tok::r_brace) && 1071 (!RootToken.Next || 1072 (RootToken.Next->is(tok::semi) && !RootToken.Next->Next))) 1073 Newlines = std::min(Newlines, 1u); 1074 if (Newlines == 0 && !RootToken.IsFirst) 1075 Newlines = 1; 1076 1077 // Insert extra new line before access specifiers. 1078 if (PreviousToken && PreviousToken->isOneOf(tok::semi, tok::r_brace) && 1079 RootToken.isAccessSpecifier() && RootToken.NewlinesBefore == 1) 1080 ++Newlines; 1081 1082 Whitespaces.replaceWhitespace( 1083 RootToken, Newlines, Indent, Indent, 1084 InPPDirective && !RootToken.HasUnescapedNewline); 1085 } 1086 1087 FormatStyle Style; 1088 Lexer &Lex; 1089 SourceManager &SourceMgr; 1090 WhitespaceManager Whitespaces; 1091 std::vector<CharSourceRange> Ranges; 1092 std::vector<AnnotatedLine> AnnotatedLines; 1093 1094 encoding::Encoding Encoding; 1095 bool BinPackInconclusiveFunctions; 1096 }; 1097 1098 } // end anonymous namespace 1099 1100 tooling::Replacements reformat(const FormatStyle &Style, Lexer &Lex, 1101 SourceManager &SourceMgr, 1102 std::vector<CharSourceRange> Ranges) { 1103 Formatter formatter(Style, Lex, SourceMgr, Ranges); 1104 return formatter.format(); 1105 } 1106 1107 tooling::Replacements reformat(const FormatStyle &Style, StringRef Code, 1108 std::vector<tooling::Range> Ranges, 1109 StringRef FileName) { 1110 FileManager Files((FileSystemOptions())); 1111 DiagnosticsEngine Diagnostics( 1112 IntrusiveRefCntPtr<DiagnosticIDs>(new DiagnosticIDs), 1113 new DiagnosticOptions); 1114 SourceManager SourceMgr(Diagnostics, Files); 1115 llvm::MemoryBuffer *Buf = llvm::MemoryBuffer::getMemBuffer(Code, FileName); 1116 const clang::FileEntry *Entry = 1117 Files.getVirtualFile(FileName, Buf->getBufferSize(), 0); 1118 SourceMgr.overrideFileContents(Entry, Buf); 1119 FileID ID = 1120 SourceMgr.createFileID(Entry, SourceLocation(), clang::SrcMgr::C_User); 1121 Lexer Lex(ID, SourceMgr.getBuffer(ID), SourceMgr, 1122 getFormattingLangOpts(Style.Standard)); 1123 SourceLocation StartOfFile = SourceMgr.getLocForStartOfFile(ID); 1124 std::vector<CharSourceRange> CharRanges; 1125 for (unsigned i = 0, e = Ranges.size(); i != e; ++i) { 1126 SourceLocation Start = StartOfFile.getLocWithOffset(Ranges[i].getOffset()); 1127 SourceLocation End = Start.getLocWithOffset(Ranges[i].getLength()); 1128 CharRanges.push_back(CharSourceRange::getCharRange(Start, End)); 1129 } 1130 return reformat(Style, Lex, SourceMgr, CharRanges); 1131 } 1132 1133 LangOptions getFormattingLangOpts(FormatStyle::LanguageStandard Standard) { 1134 LangOptions LangOpts; 1135 LangOpts.CPlusPlus = 1; 1136 LangOpts.CPlusPlus11 = Standard == FormatStyle::LS_Cpp03 ? 0 : 1; 1137 LangOpts.LineComment = 1; 1138 LangOpts.Bool = 1; 1139 LangOpts.ObjC1 = 1; 1140 LangOpts.ObjC2 = 1; 1141 return LangOpts; 1142 } 1143 1144 } // namespace format 1145 } // namespace clang 1146